41297 Chap.5 Percentiles, Quartiles and the Five-Number Summary
Percentiles, Quartiles and the Five-Number Summary
Beyond the centre
A single average hides where the other values sit. Fractiles fill that gap: quartiles cut sorted data into four equal parts, deciles into ten and percentiles into a hundred, so each one names a value below which a stated share of observations falls.
One rule for every cut point
The lecture uses a location index, p divided by 100 and multiplied by n.
A fractional index is rounded up to the next position; a whole-number index means averaging that position and the next. The same rule gives quartiles, the median and any percentile.
Summaries that resist outliers
The minimum, first quartile, median, third quartile and maximum form the five-number summary, the skeleton of a box plot.
The interquartile range, Q3 minus Q1, measures the spread of the middle half and is unaffected by extreme values, unlike the range.
What this chapter covers
- 01
Quartiles, deciles and percentiles
- 02
The location index rule
- 03
Rounding up versus averaging two positions
- 04
Five-number summary and box plots
- 05
Interquartile range against range
Worked example · free
Five-number summary and IQR of call times
- 1Q1: index 0.25 × 12 = 3 is whole, so average the 3rd and 4th values: (6 + 6) ÷ 2 = 6.
- 1Median: index 0.5 × 12 = 6, so average the 6th and 7th values: (8 + 9) ÷ 2 = 8.5.
- 1Q3: index 0.75 × 12 = 9, so average the 9th and 10th values: (11 + 12) ÷ 2 = 11.5.
- 1Minimum 4 and maximum 22 complete the summary: 4, 6, 8.5, 11.5, 22.
- 1IQR = 11.5 − 6 = 5.5 minutes against a range of 18, inflated by the single 22-minute call.
Key terms
- Percentile
- A cut point below which a given percentage of the sorted observations lie.
- Quartile
- One of three cut points dividing sorted data into four parts holding a quarter of the values each.
- Location index
- The position p divided by 100 times n that tells you where a percentile sits in sorted data.
- Five-number summary
- The minimum, first quartile, median, third quartile and maximum reported together.
Percentiles, Quartiles and the Five-Number Summary FAQ
What does the 90th percentile actually tell a manager?
It is the value that 90% of cases do not exceed. For delivery times, a 90th percentile of 43 minutes means nine in ten parcels arrive within 43 minutes, which is a far more useful service promise than the average, because the average says nothing about how long the slowest deliveries take.
Do I round the location index up or down?
Up, whenever it is fractional. An index of 4.5 or 4.25 sends you to the 5th sorted value. Only when the index is a whole number do you average two values, the one at that position and the one after it. Rounding down is the most frequent slip and shifts every quartile one place too low.
Why do spreadsheet quartiles sometimes differ from my hand calculation?
Software packages use several conventions for interpolating between positions, so small datasets can give slightly different quartiles depending on the function chosen. For seminar work and the report, follow the method taught in the lecture, state that you used it, and treat small differences from software output as a matter of convention rather than an error.
Why is the interquartile range preferred to the range for skewed data?
The range depends only on the two most extreme observations, so one unusual case can double it. The interquartile range discards the lowest and highest quarters and measures how far apart the middle half sits, which makes it stable when a dataset has a long tail or a few outliers.
Assessment move
Practise the location index until it is automatic: take ten random columns of different lengths from your dataset and, for each, write the index for Q1, the median, Q3 and the 90th percentile before reading off any value. Mix even and odd lengths so you meet both branches of the rule. Then draw a quick box plot by hand from the five-number summary and describe it in two sentences, one on the middle half and one on the tails.
Compare your quartiles with spreadsheet output and note any convention differences rather than assuming you are wrong. Keep the worked sheets, because the report will need these summaries for your chosen variables.
Working through Percentiles, Quartiles and the Five-Number Summary in 41297? Sia is AskSia’s AI Statistics tutor — ask any 41297 Percentiles, Quartiles and the Five-Number Summary question and get a clear, step-by-step explanation grounded in how 41297 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.